study in England - Europe PMC

7 downloads 0 Views 1MB Size Report
G Bickler. Bromley Health,. Department of Public. Health Medicine,. Hayes, Kent BR2 7EH. J Hunter. Correspondence to: Dr S Sutton. Accepted for publication.
413

3rournal of Epidemiology and Community Health 1995;49:413-418

Does routine screening for breast cancer raise anxiety? Results from a three wave prospective study in England Stephen Sutton, Guitta Saidi, Graham Bickler, Janet Hunter

ICRF Health Behaviour Unit, Institute of Psychiatry, 4 Windsor Walk, London SE5 8AF S Sutton G Saidi

South East London Health Authority, 1 Lower Marsh, London SE1 7RJ G Bickler

Bromley Health, Department of Public Health Medicine, Hayes, Kent BR2 7EH J Hunter Correspondence to: Dr S Sutton. Accepted for publication February 1995

Abstract Objective - To investigate whether mammography raises anxiety in routinely screened women who receive a negative result. Design - Prospective design in which women completed questionnaires at three key points in the breast screening process: at baseline (before being sent their invitation for breast screening), at the screening clinic immediately before or after screening, and at follow up, about nine months after baseline. Information was obtained from non-attenders as well as from attenders. Setting - Bromley District Health Authority, served by the South East London Breast Screening Service. Participants - Two overlapping samples were used. Sample A comprised 1500 women aged 50-64 who were due to be called for first round screening at a mobile screening unit. Altogether 1021 (68%) returned a usable questionnaire and 795 of these (78%) also provided adequate information at nine month follow up: there were 695 attenders (including 24 women who received false positive results) and 100 non-attenders. Sample B consisted of 868 women who attended the screening unit in a three month period, 732 (84%) of whom provided adequate data. A total of 306 attenders (including 10 who received false positive results) occurred in both samples and provided adequate information on all occasions. The main analyses were based on these 306 women plus the 100 nonattenders. The analysis of retrospective anxiety took advantage of the larger sample size of 695 attenders. Main results - On average, the women were not unduly anxious at any of the three points in the screening process. Among attenders, there was no difference between anxiety levels immediately before and immediately after screening. Anxiety was lowest at the clinic and highest at baseline but the changes were very small in absolute terms. Anxiety did not predict attendance: there were no differences in anxiety levels between attenders and non-attenders at baseline. As expected, women who received false positive results recalled feeling extremely anxious after they had received the referral letter but their retrospective anxiety was also higher than in the negative screenees at earlier stages in the breast

screening process. They also reported having experienced more pain and discomfort during the x ray. Conclusions - Anxiety does not seem to be an important problem in routinely screened women who receive a negative result. This finding is very reassuring in relation to a major criticism of breast screening programmes. Thus, apart from maintaining current procedures such as keeping waiting times to a minimum, there seems to be no need to introduce special anxiety reducing interventions into the national programme. On the other hand, the findings for women who received false positive results suggest that there are aspects of the experience of being recalled

for assessment after an abnormal mammogram that warrant further attention. The relationship between contemporaneous and retrospective anxiety should also be studied.

(J Epidemiol Community Health 1995;49:413-418) There is increasing interest in assessing the potential psychological costs of screening and other medical procedures.l" It has been suggested that the anxiety aroused by breast screening may significantly offset the benefits of reduced mortality.5-7 Although critics of mass mammography emphasise the possible adverse psychological impact of receiving false positive results, it is important to establish whether routine screening for breast cancer also raises anxiety in the vast majority (>90%) of women who receive a negative result. This was the main aim of the present study, which was conducted within the UK National Health Service Breast Screening Programme. A secondary aim of the study was to compare attenders and nonattenders with regard to anxiety in order to investigate the notion that screening programmes tend to attract women who are characteristically more (or less) anxious than average. A number of studies have measured anxiety or psychiatric morbidity in women who receive a negative result after mammography.8-"1 These and other relevant studies'2-'4 have been reviewed elsewhere.4 Their results suggest that in general women who receive a negative result are not made anxious and may be reassured by their experience of participating in breast screening. None of the studies, however, included a baseline measure of anxiety, or more

414

Sutton, Saidi, Bickler, Hunter Sample A 1500 randomly selected from 2 screening batches and sent baseline questionnaire

1021 (68%) provided adequate data

1012 were sent follow up questionnaire

Sample B 795 (79%) were provided adequate data at both waves: 695 attenders 100 non-attenders

868 attenders approached at screening clinic

Overlap sample n = 306 attenders

732 attenders (84%) provided adequate data

Fig 1 Flow diagram showing the design of the study.

than two waves of measurement, so it is difficult to exclude the possibility that attending for screening is associated with a real increase in anxiety which may persist for several months. Furthermore, few studies have assessed anxiety in those who fail to attend and thus have a direct bearing on the issue of whether breast screening programmes tend to attract women who are psychologically more (or less) healthy than average. In the study reported here, a cohort of women was followed through the screening process, with anxiety measures at several key points at baseline (before invitation); at the screening clinic, either immediately before or immediately after screening; and at follow up, about nine months after baseline. Retrospective as well as contemporaneous measures of anxiety were used. Information was obtained from non-attenders as well as attenders. Although not the main focus of the present study, our sample included some women who were recalled for investigation after a positive breast screen but subsequently received a normal result. The results from this subgroup of subjects with "false positives" are also reported.

Methods SAMPLES

Two overlapping samples, A and B, were used. Sample A consisted of 1500 women aged 50-64 living in Bromley District Health Authority who were due to be called for first round screening at the Bromley Mobile Breast Screening Unit. They were drawn randomly from two consecutive screening batches. Because of the way the screening batches were generated (by month and year of birth), the sample should be representative of women aged 50-64 who are registered with general practitioners in Bromley. Sample B consisted of 868 women who attended the mobile unit for mammography between November 1991 and February 1992.

sample A were sent a questionnaire through the post. A second questionnaire was sent to non-responders two to three weeks after the first. A follow up questionnaire was sent about nine months later, again with a second mailing to non-responders. Nine women who had been found to have breast cancer or were still undergoing investigation were excluded from the follow up. However, 35 women ("false positives") who were recalled for investigation after a positive screen but were subsequently declared normal were included. Information on attendance was obtained from the screening office. Sample A thus provided baseline (pre-invitation) and follow up measures of anxiety for both attenders and non-attenders. In order to obtain measures of anxiety from women (including those in sample A) while attending the screening clinic, four half-day clinic sessions per week were sampled at random over the three month period during which most women in sample A would be expected to attend for screening. Women attending these sessions were approached by an experienced female interviewer (GS) and asked to complete a short questionnaire. Sessions were designated "before" or "after" sessions in a random fashion. In "before" sessions, the questionnaires were administered while the women were waiting to be screened; in the "after" sessions, the questionnaires were administered immediately after they had been screened. Given the time and space constraints imposed by the environment of a mobile clinic, it would not have been practicable to obtain before and after measures from the same women. In addition, we wished to avoid creating a demand effect; women completing the questionnaire for the second time in the space of a few minutes may have felt subtle pressure to indicate that their anxiety had decreased. A total of 868 women were approached at the screening clinic (sample B). This procedure succeeded in its aim of producing a sizeable overlap between sample A and sample B. The design of the study is shown in Figure 1. MEASURES

Baseline questionnaire The baseline postal questionnaire included the following measures:

(1)

The

Spielberger state-trait anxiety inventory

(STAI, form-Y).15 This is a widely used, 40 item scale. The state scale was designed to assess transient states of anxiety (as might be experienced for example by a student about to sit an examination). In contrast, the trait scale is intended to assess a predisposition towards anxiety or the extent to which someone is characteristically anxious.

(2) The general health questionnaire (GHQ) anxiety subscale. The seven item anxiety subscale from the GHQt6 was also included.

DESIGN AND PROCEDURE

One to three months before the invitations for screening were due to be sent out, women in

(3) Sociodemographic factors. Items on age, marital status, education (whether respondent

415

Routine breast cancer screening and anxiety

had any educational qualifications), employment (whether respondent did any paid work), housing tenure (an indicator of social class) were included. Clinic questionnaire Because of the time restriction, only the state anxiety questionnaire was used at the screening clinic.

were repeated, controlling for each ofthe demographic factors in turn. Since the main results and conclusions were unchanged, only the original, unadjusted results are reported. Because the anxiety scores were highly skewed, the main results were checked - firstly, by using log transformed scores and secondly, by using nonparametric tests. All analyses were done using SPSSPC 4. 1. Results

Follow up questionnaire The nine month follow up questionnaire repeated the anxiety scales from the baseline questionnaire. Among those who had attended for screening, retrospective ratings of anxiety, pain, and discomfort were also obtained. FINAL SAMPLES

Sample A Altogether 1021 out of 1500 women (68%) returned a usable baseline questionnaire, and 795 (78%) of these (53% of 1500) provided adequate data at baseline and follow up. There were 695 attenders (including 24 with false positive results) and 100 non-attenders.

Sample B Of the 868 women approached at the screening clinic, 732 (84%) provided adequate data. Common reasons for women refusing to fill out the questionnaire or failing to finish it were lack of time and not having their reading glasses with them.

Overlap between A and B A total of 306 attenders (including 10 women who received false positive results) occurred in both samples and provided adequate information at all stages. The main analyses of change in anxiety were based on these 306 attenders plus the 100 non-attenders from sample A who provided adequate information at baseline and follow up. In this subsample of 406 women, the mean age was 58 years, 76% were married or living as married (17% divorced, separated or widowed; 7% single), 50% had an educational qualification, 53% were in paid employment, and 91% owned or were buying their home. The subsample of 406 was very similar in both demographic characteristics and initial anxiety levels to the full sample of 1021 who returned a usable baseline questionnaire; thus there was no evidence of strong attrition bias. The analysis of retrospective anxiety (which was assessed on the follow up questionnaire) used a larger sample size of 695 attenders who provided usable information at this stage. STATISTICAL ANALYSIS

The main procedures employed in the analysis were product-moment correlations, independent and paired t tests, and repeated measures analysis of variance (ANOVA). Two sided significance tests and an alpha level of *05 were used throughout. The main analyses

COMPARISON OF STATE ANXIETY LEVELS IMMEDIATELY BEFORE AND AFTER SCREENING

There was no significant difference between state anxiety levels immediately before and after screening. Some women, however, filled in the questionnaire after instead of before screening. This could have biased the before-after comparison if these women had had relatively high anxiety levels. To check on this, the analysis was repeated after excluding the 36 women for whom it had not been possible to adhere to the random allocation procedure. The results were unchanged; in particular, there was still no suggestion of a before-after difference in state anxiety. In subsequent analyses of state anxiety, the distinction between before and after screening is not retained; we simply use a single measure to represent the level of state anxiety experienced at the screening clinic. CHANGES IN ANXIETY LEVELS DURING THE SCREENING PROCESS

Mean anxiety scores at different stages in the screening process are shown in figure 2 for * o * o A I

Trait - attenders Trait - non-attenders State-attenders State - non-attenders GHQ-attenders GHQ - non-attenders

2rF 41

a) x

c

n U,

0

a) c co

1

0

0. a

E 0

0 v

1~

2

3

Stage in screening progress Figure 2 Mean anxiety scores for attenders and nonattenders at baseline (1), at clinic (2), and at follow up (3). All three anxiety questionnaires used four point scales and were scored conventionally as follows: State anxiety: 1= "not at all", 2 = "somewhat", 3= "moderately so", 4= "very much so". Trait anxiety: 1= "almost never", 2 = "sometimes", 3= "often ", 4 = "always". GHQ anxiety: 0= "not at all", 1= "no more than usual", 2 = "rather more than usual", 3 = "much more than usual". Note: Although it is common practice to report mean total scores on the state and trait anxiety scales (theoretical range 20-80), we report mean mean scores (theoretical range 1-4) because some participants had missing data on some items.

416

Sutton, Saidi, Bickler, Hunter

>3 a) Cx

Negatives A False positives 0

A

a) >

2

0

a)

a) U

0

1

-a,

2 3 4 5 Stages in screening process

6

Figure 3 Mean retrospective anxiety for the two groups of attenders at each of six stages in the breast screening process. Key to stages: I = "when you received the invitation letter"; 2 -"at the screening centre while you were waiting to have the x ray"; 3 "at the screening centre after you had had the x

p< 001). However, the expected pattern of change (an increase between baseline and clinic followed by a reduction at follow up) did not emerge. On the contrary, women were least anxious, on average, when they were at the screening clinic. The changes were again quite small in absolute terms. Excluding the 10 women with false positive results did not change the results reported above. Repeated measures ANOVAs comparing the false positives with the other attenders (that is, the negative screenees) showed no differences between the two groups in anxiety levels or in the pattern of change in anxiety over time. ANALYSIS OF RETROSPECTIVE ANXIETY

On the follow up questionnaire, those who had attended for screening were asked to rate how 4 "in the period between going for screening and anxious they felt at each of six stages in the receiving your results screening process. A simple three point scale was 5 "after you'd opened and read the results letter"; used (1 not at all anxious, 2 =a bit anxious, 6 = "now". 3 =very anxious). As figure 3 shows, retrospective anxiety varied considerably across stages and showed a different pattern in the attenders and non-attenders separately. The two groups. (Repeated measures ANOVA commeans for state and trait anxiety all fell within paring the false positives with the negative the range 1-65-1-95. Although somewhat screenees yielded significant effects for group higher than published norms for women in this (F(1, 665)=28 6, p< 001), for time (F(5, age group, they are fairly typical of values 661)= 143 4, p< 001), and for the group by obtained from working adults tested under un- time interaction (F(5, 661)=63-7, p< 001).) stressful conditions.'5 (Psychiatric and general As would be expected, compared with the negmedical and surgical patients typically have ative screenees, the women with false positive mean scores >2.0). State anxiety was conresults recalled feeling extremely anxious at sistently lower than trait anxiety, which suggests stage 5 (after they had opened and read the that the women were not unduly stressed.'5 results letter): mean (SD) 2 85 ( 37) and 1 16 Repeated measures ANOVAs conducted on ( 36); t(665)=21 1, p< 001. What is of pareach of the three anxiety scales using the base- ticular interest, however, is that retrospective line and follow up data showed no significant anxiety was also higher in this group than in differences between attenders and non-at- the negative screenees at stage 3 of the process, tenders (no main effect for group). Neither that is when they were at the screening clinic after being screened (mean (SD) 1 60 ( 68) and state nor trait anxiety showed a significant change from baseline to follow up. By contrast, 1-36 ( 52) respectively; t(665)=2 01, p< 05). GHQ anxiety showed a significant reduction The difference in anxiety at stage 4 (in the period between going for screening and reover time: F(1, 404)=11 4, p< 001. The change was very small, however, in absolute ceiving their results) was also close to sigterms (equivalent to less than one tenth of a nificance (mean (SD) 1 95 ( 09) and 1 70 (-57) unit on the four-point scale). None of the group respectively; t(665)= 1-93, p= 054). The (attenders v non-attenders) analyses in relation women with false positive results also reported to time interactions was significant. Thus there having experienced significantly more pain and discomfort during the x ray than women with was no evidence that baseline anxiety predicted attendance and little evidence that attendance negative results (mean (SD) for pain on a fourpoint scale from 1 = no pain to 4 = severe pain influenced anxiety, at least in the longer term. Correlations between individuals' anxiety were 2 04 (0 98) and 1 62 (0 70) respectively; t(680) 2.80, p< 005). Thus, for this subgroup scores at baseline and follow up indicated considerable stability over time, particularly for of women, the memory of the experience of going for screening seemed to have been coltrait anxiety (r=0 79, p< 001). Additional analyses incorporating demo- oured by their later experiences. Retrospective anxiety at stage 6 ("now") graphic factors did not change the results reported above but showed a consistent effect of correlated 27--28 (p